Results from the Green
House Evaluation
in Tupelo, MS
Rosalie A. Kane, Lois J. Cutler, Terry Lum &
Amanda Yu
University of Minnesota,
funded by the Commonwealth Fund.
Academy Health Annual Meeting,
June 26, 2005
Green House Summarized



Radical culture change for NH-level care
Environment, organization & philosophy
Main elements
– 10 elders live in self-contained houses with private
rooms & baths
– meals cooked in GH kitchen in presence of elders
– care from CNA-level resident assistants (Shahbazim)
who cook, do housekeeping, personal care, laundry, &
facilitate elder development
– Shahbazim do not report to nursing
– All professionals (RNs, MD, SW, RT, PT, OT, etc)
comprise clinical support teams that visit GHs
Green House description, cont’d

A group of GHs are licensed as a NF
– share administrative support & clinical support teams




GH receives same reimbursement as any
Medicaid NH
GH meets all standards of licensure &
certification
Emphasis on quality of life for elders (quality of
care is a given, but health & safety goals do not
dominate model)
GH idea developed by Bill Thomas. (more info at
http://thegreenhouseproject.com/)
Tupelo Green Houses





Sponsor: Cedars Health Care Center, a 140-bed
traditional NH on campus of Traceway
Retirement Community, owned by Methodist
Senior Services of Mississippi
Line staff trained intensively in GH model
Elders moved from Cedars to 4 10-person GHs
built in residential area of campus in May 2003
2 of the GHs were populated by the former
residents of the locked dementia care unit
Admission from Cedars to fill GH vacancies
Photo by: The Green House Project
Waterville, NY
Study components




Implementation study
Environmental study
Business case study
Longitudinal study of
outcomes
– quantitative and qualitative
Experimental design

Quasi-experiment with 2 comparison groups
– sample of residents remaining at Cedars NH
– sample from Trinity Health Care, a NH of same owner
in nearby city

4 waves of data collection
– May-June 2003 (pre-move data)
– 3 more times at 6 month intervals

Respondents
– Residents
– Primary family caregiver
– All Shahbazim and CNAs
Hypotheses

Residents:
– QOL,  satisfaction,  social engagement
– health no worse than in conventional “medical model”

Family caregivers
–  engaged with residents,  satisfied,  burden

Front-line staff
–  knowledgeable about residents,  confident about
affecting QOL,  intrinsic and extrinsic job
satisfaction,  likelihood of staying in job

MDS-derived QIs expected to be as good or
better than conventional “medical model”
Analysis
Multivariate analysis
 4 waves pooled and time (i.e. wave)
entered as a variable
 random effects panel analysis methods
 Case mix adjustment used

Other methods (not presented here)
include comparison of change scores
over time
Selection issues

GH elders were similar in disability to
those who remained at Cedars.
– No statistically significant differences in
gender, ADLs, levels of behavior problems,
LOS from admission
– Cedars residents were slightly more
depressed, and cognitively impaired
(p<.05).
Study Samples Similar at
Baseline
Cedars
Green
House
Trinity
Sample size (N)
40
40
40
Female (in %)
88%
80%
75%
Age (in years, mean)
87
81
89
Average self-reported health (1-poor, 5-excellent, mean)
2.4
2.6
2.5
Cognitive impairment from MDS (0-5)
3.7*
2.8
3.2
ADL Difficulty from MDS
8.6
7.0
8.4
Depression from MDS
0.9*
0.6
0.3
Behavioral problem from MDS
1.2
0.7
1.5
LOS (in days)
1193
682
1108
*<.05, **<.01, ***<.001
Resident Outcomes

Compared to the 2 control settings,
GH residents reported a better quality
of life and greater satisfaction.
Resident QOL
Model xttobit
Quality of Life Subscales
Cedars
Coef. (S.E.)
Trinity
z-value
Coef. (S.E.)
z-value
Comfort scale
-0.003 (0.057)
-0.05
-0.015 (0.067)
-0.22
Functional competence scale
-0.158 (0.147)
-1.08
0.020 (0.167)
0.12
Privacy scale
-0.580 (0.113)
-5.12***
-0.150 (0.130)
-1.15
Dignity scale
-0.562 (0.108)
-5.20***
-0.322 (0.124)
-2.60**
Meaningful activity scale
-0.185 (0.070)
-2.66**
-0.001 (0.080)
-0.01
Relationship scale
-0.331 (0.099)
-3.36**
-0.082 (0.113)
-0.73
Autonomy scale
-0.280 (0.104)
-2.68**
-0.132 (0.121)
-1.09
Food enjoyment scale
-0.625 (0.136)
-4.59***
-0.501 (0.157)
-3.20**
Spiritual well-being scale
-0.129 (0.117)
-1.10
0.268 (0.135)
1.99*
Security scale
-0.050 (0.044)
-1.13
0.009 (0.051)
0.18
Individuality scale
-0.374 (0.093)
-4.00***
-0.076 (0.108)
-0.71
Controlled for: wave, gender, age, self-reported health, length of stay, ADL (from MDS), and
cognition (from MDS) *p<.05, **p<.01, ***p<.001
Resident QOL – single items
Trinity
Cedars
Quality of Life
Coef. (S.E.)
z-value
Coef. (S.E.)
z-value
Physically comfort CMF
-0.427 (0.148)
-2.88**
-0.383 (0.169)
-2.27*
Doing as much for yourself FC
-0.438 (0.140)
-3.12**
-0.257 (0.159)
-1.61
Having privacy PRI
-0.815 (0.152)
-5.35***
-0.444 (0.171)
-2.60**
Having choice & control AUT
-0.500 (0.132)
-3.80***
-0.234 (0.151)
-1.68
Dignity is respected DIG
-0.319 (0.161)
-1.99*
-0.134 (0.183)
-0.73
Interesting things to do MA
-0.220 (0.137)
-1.61
0.228 (0.156)
1.46
Enjoying foods ENJ
-0.477 (0.163)
-2.92**
-0.292 (0.187)
-1.56
Interest & preferences IND
-0.303 (0.151)
-2.01*
-0.081 (0.172)
-0.47
Good friendship REL
-0.436 (0.155)
-2.82**
-0.181 (0.176)
-1.03
Feel safe SEC
-0.123 (0.165)
-0.75
0.016 (0.189)
0.09
Spiritual and religion needs
SWB
-0.294 (0.149)
-1.98*
0.250 (0.171)
1.46
Overall quality of life
-0.535 (0.150)
-3.57***
-0.366 (0.172)
-2.13*
QOL-sum of previous items
except QOL1
-3.075 (0.853)
-3.60***
-0.607 (0.982)
-0.62
Controlled for: wave, gender, age, self-reported health, length of stay, adl (from MDS), and cognitive (from MDS)
*p<.05, **p<.01, ***p<.001
Residents’ Satisfaction
Model xttobit
Overall Satisfaction
Cedars
Coef. (S.E.)
Trinity
z-value
Coef. (S.E.)
z-value
Place to live
-0.985 (0.144)
-6.86***
-0.742 (0.162)
-4.61***
Place to get care
-0.717 (0.139)
-5.16***
-0.390 (0.159)
-2.46*
Recommend
-0.932 (0.164)
-5.67***
-0.450 (0.184)
-2.45*
Controlled for: wave, gender, age, self-reported health, length of stay, ADL (from MDS), and cognitive
(from MDS) *p<.05, **p<.01, ***p<.001
Family Outcomes


Compared to the 2 control settings,
GH family members reported greater
satisfaction with their relative's care
and life.
They also reported greater satisfaction
with how they as family members
were treated.
Family outcomes
Model xttobit
Trinity
Cedars
Coeff.
z-value
Coef.
z-value
Family’s satisfaction in:
Food, dining ambiance,
housekeeping, disability-friendly
environment (4 items)
-3.801
-8.25***
Social activity (4 items)
-0.855
-1.67
Room and bath (3 items)
-5.255
-10.71***
-2.879
-4.97***
-4.369
-5.23***
-3.103
-3.01**
-4.564
-7.20***
-1.892
-2.44*
-6.213
-5.91***
-2.871
-2.16*
Help give to resident
2.869
3.38**
2.616
2.20*
Objective burden
1.530
1.75
1.750
1.61
Subjective burden
0.416
0.46
-0.224
-0.19
Autonomy & Relationships with
staff (6 items)
Nursing care (5 items)
Experience as family (7 items)
Controlled for: wave and gender
*<.05, **<.01, ***<.001
-2.487
0.954
-4.26***
1.49
Staff outcomes




Compared to the 2 control settings,
staff felt more empowered to assist
residents.
They knew residents better
They experienced greater intrinsic and
extrinsic job satisfaction
They were more likely to plan to
remain in the job.
Staff outcomes
Model xttobit
Trinity
Cedars
Staff Measure
Staff Knowledge about residents (5
items)
Coef.
z-value
Coef.
z-value
-3.216
-6.04***
-0.972
-1.45
-10.849
-4.77***
-9.884
-3.55***
-3.945
-3.99***
-3.137
-2.56*
-1.393
-1.56
-2.728
-2.50*
-1.184
-1.84
-2.019
-2.63**
Empowerment: Belief that she can influence:#
Resident quality of life (12 items)
Family experience (2 items)
Resident’s physical health & functioning
(3 items)
Resident’s emotional well-being (2 items)
Controlled for: wave and gender *<.05, **<.01, ***<.001
# each item rated from 10 (maximum influence to 1 minimum influence
Staff Appraisal of Jobs
Trinity
Cedars
Ratings of job characteristics
Sense of belonging (3 items)
-1.989
-5.84***
-1.143
-2.70**
Sense of competence (3 items)
-0.333
-1.39
-0.032
-0.11
Coworker support (3 items)
-0.642
-2.16*
0.289
0.79
Depersonalization (4 items)
0.871
3.34**
0.249
0.78
Information exchange (4 items)
-2.219
-6.02***
-1.394
-2.69**
Management support (3 items)
-1.853
-5.86***
-0.303
-0.72
Supervisor support (3 items)
-2.157
-6.22***
-0.987
-2.15*
Training (3 items)
-1.384
-5.17***
-1.008
-2.91**
Intrinsic satisfaction (15 items)
-4.726
-6.60***
-3.532
-3.69***
Extrinsic Satisfaction (4 items)
-2.855
-6.78***
-2.422
-4.26***
-2.927
-5.23***
-2.841
-3.90***
Job satisfaction
Intent to stay in job (4 items)
Controlled for: wave and gender *<.05, **<.01, ***<.001
Quality Indicators


MDS-based QI analyses over a 2 year period
showed either no difference in QIs or
statistically significant advantages for GH.
Compared to the control settings, the GH
showed less ADL decline, less prevalence of
depression, less incontinence without a
toileting plan, and less use of antipsychotics without a relevant diagnosis.
Quality Indicators
Cedar
(N=109)
Variable
Trinity
(N=69)
OR
Std. Dev.
OR
Std. Dev.
qi1 Incidence of New Fracture
0.38
0.39
0.54
0.51
qi2 Prevalence of fall
1.99
0.85
1.96
0.89
qi3 Prevalence of behavioral Symptom
0.55
0.30
1.79
1.06
qi4 Prevalence of depression
2.65*
1.16
3.34*
1.56
qi5 Prevalence depression without antidepressant
1.68
1.05
1.57
1.09
qi6 Use of 9 or more medications
0.45
0.34
0.26
0.21
qi7 incidence of cognitive impairment
1.97
1.36
0.44
0.33
qi8 Prevalence of incontinence
1.97
1.36
0.44
0.33
qi9 Prevalence of incontinence without toilet plan
11.0*
13.1
62.8**
95.9
qi10 Prevalence of indwelling catheters
1.06
1.17
1.42
1.91
qi11 Prevalence of fecal impaction
n/a
n/a
n/a
n/a
qi12 Prevalence of urinary tract infections
1.06
0.43
2.37*
0.67
Quality Indicators
Cedar
(N=109)
Variables
Trinity
(N=69)
qi13 Prevalence of weight loss
0.91
0.29
0.88
0.30
qi14 Prevalence of tube feeding
0.04
0.07
0.002
0.007
qi15 Prevalence of dehydration
n/a
n/a
n/a
n/a
qi16 Prevalence of bedfast residents
3.94
3.28
0.18
0.20
qi17 Inc. of decline of late loss ADL
3.54***
1.10
5.72***
1.98
qi18 Inc. of decline of ROM
0.82
0.37
0.43
0.23
qi19 Prevalence of antipsychotic use
5.06**
3.05
0.31
0.24
qi20 Prevalence of anti-anxiety / hypnotic use
0.85
0.57
3.75
2.77
qi21 Prevalence of hypnotic use > 2 times
3.45
4.50
4.21
8.17
qi22 Prevalence of daily physical restraints
0.97
7.55
0.57
0.48
qi23 Prevalence of little or no activity
12.26***
8.41
0.73
0.54
qi24 Prevalence of stage 1-4 pressure ulcer
2.59
1.95
2.99
2.36
* p<.05; ** p<.01; *** p<.001
Control variables: age, gender, ADLs, cognition, and wave, except in qi17 where ADL was not used.
Conclusions





Outcome findings robust in support of GH
for residents, family, and staff compared to
Cedars , Trinity, or both
Almost no negative findings
Staff findings striking, suggesting staff
empowerment is possibly a vehicle for
resident outcomes
Sponsor implementing 100% GHs
Many other GHs under development
Next steps for present study

For first 4 GHs
– More analysis, including open-ended
items
– Analysis of 100+ exit interviews from
former staff and from families of
deceased residents from GH and
comparison settings
– Environmental analyses
– Comparisons of specific GHs
– Business case data
– Implementation issues
Studies of Tupelo expansion

Full scale 100% GH will permit:
– insight on managing a larger collection of
GHs & utilizing centralized staff
– examining admission direct from
community
– Studying integration of new elders & staff
into GHs on larger scale
– Comparing 10-person & 12-person GHs
– MDS QIs for full GH
Thank you



To the Commonwealth Fund
To all the respondents in Mississippi
And to all of you for listening
For more information, contact:



kanex002@umn.edu
cutle001@umn.edu
tlum@umn.edu